X-Git-Url: https://git.cweiske.de/phinde.git/blobdiff_plain/d8c39f2b0571b9734259b2f9dc218eed24412332..a1c8309e4b3d8d5468c2defbc8bbae95633aff90:/src/phinde/LinkExtractor/Html.php

diff --git a/src/phinde/LinkExtractor/Html.php b/src/phinde/LinkExtractor/Html.php
index a6fa8ef..299ed91 100644
--- a/src/phinde/LinkExtractor/Html.php
+++ b/src/phinde/LinkExtractor/Html.php
@@ -8,7 +8,7 @@ class Html
 {
     public function extract(\HTTP_Request2_Response $res)
     {
-        $url = $res->getEffectiveUrl();
+        $url = Helper::removeAnchor($res->getEffectiveUrl());
 
         $linkInfos = array();
 
@@ -22,6 +22,13 @@ class Html
 
         $dx = new \DOMXPath($doc);
 
+        $xbase = $dx->evaluate('/html/head/base[@href]')->item(0);
+        if ($xbase) {
+            $base = $base->resolve(
+                $xbase->attributes->getNamedItem('href')->textContent
+            );
+        }
+
         $meta = $dx->evaluate('/html/head/meta[@name="robots" and @content]')
             ->item(0);
         if ($meta) {
@@ -54,7 +61,7 @@ class Html
                     }
                 }
             }
-            if ($href == '' || $href{0} == '#') {
+            if ($href == '' || $href[0] == '#') {
                 //link on this page
                 continue;
             }